An Easily Accessible, Semi-Automated Approach to Creating Personalized Normative Feedback and Risk Feedback Graphics

Abstract

Objective:

Personalized normative feedback interventions show efficacy in reducing health risk behaviors (e.g., alcohol use, sexual aggression). However, complex personalized normative feedback interventions may require manual methods of inputting participant data into graphics, which introduces error, and automated approaches require substantial technical costs and funding and may limit the types of feedback that can be provided.

Method:

To make personalized normative feedback more accessible, we outline a method of using easily accessible software programs, including IBM Statistical Package for the Social Sciences (SPSS), Microsoft Excel, and Microsoft PowerPoint, to create and display complex personalized normative feedback graphics. We also describe methods through which personalized normative feedback graphics can be created within a larger preventive intervention for alcohol and sexual assault in college men.

Results:

We first provide step-by-step instructions for collecting data and then creating semi-automated syntax files within SPSS and Excel to merge participant data into complex personalized normative feedback graphics in Excel. To do so, we append annotated syntax in text and supplemental material. Next, we outline the process of creating risk feedback graphics, whereby individual items or the exact wording of items are displayed back to the participant. Finally, we provide guidance regarding the process of translating graphics from Excel for viewing via PowerPoint without having to manually update PowerPoint slides for each presentation.

Conclusions:

Via the described syntax and graphic generation, researchers are then able to create semi-automated personalized normative feedback and risk feedback graphics. This tutorial may help increase the dissemination of complex personalized normative feedback interventions.

Personalized normative feedback interventions have become a commonly used approach for targeting risk behavior in college students and young adults. Using a social norms perspective, personalized normative feedback displays an individual's typical behavior alongside the individual's perceptions of others' behavior and others' actual behavior (Lewis & Neighbors, 2006). These interventions are thought to enact behavior change by highlighting discrepancies between (a) the perceptions of peer behavior and actual peer behavior and (b) an individual's own behavior and that of their peers.

Studies show that personalized normative feedback interventions are efficacious in reducing heavy drinking and related problems, including cannabis use, gambling, and risky sexual behavior (e.g., Borsari & Carey, 2000; Kilwein et al., 2017; Neighbors et al., 2004; Orchowski et al., 2018; Peter et al., 2019; Testa et al., 2020). Brief motivational interventions that use personalized normative feedback also commonly provide individualized risk feedback regarding specific attitudes or behaviors an individual endorses, typically at a specific level (e.g., number of occasions someone binge drinks; Doumas et al., 2010; Walters et al., 2007). This combination of personalized normative feedback and personalized risk feedback has demonstrated efficacy in decreasing risk for engagement in health risk behaviors (e.g., Walters et al., 2007).

Although personalized normative feedback and risk feedback are widely used and empirically supported interventions, and countless studies report on personalized normative feedback interventions, few discuss methods for creating and implementing personalized normative feedback/risk feedback. One method of creating personalized graphics is to use an online platform, such as Qualtrics, in which a participant responds to a question (or series of questions), and then their answers are looped into later questions and presented back to them. Although this can be an efficient and effective way to produce personalized normative feedback content, such methods become quite complex when constructs of interest require data transformations, such as reverse-scoring of items, sum scoring of items, or averaging of items to create meaningful constructs (e.g., rape-supportive attitudes, sociosexual attitudes, positive alcohol expectancies, alcohol-related problems sum) or random selection of items that a participant endorsed (e.g., expectancies, negative alcohol consequences). It may be particularly important to display a random selection of items to truly test if displaying this feedback leads to behavior change across individuals who may endorse different items. Similarly, using a single program like Qualtrics may not be possible in studies that incorporate other aspects in addition to personalized feedback, such as motivational interviewing, cognitive training, and/or verbal goal setting (e.g., Magill et al., 2022; Miller, 1983; Treat et al., 2016).

Thus, in many cases, complex personalized feedback within the context of larger intervention studies requires personnel to manually (or partially manually) input data to create personalized normative feedback/risk feedback figures. Although cost-effective, this creates a time burden for the research team and room for human error when inputting information. In other cases, grant-funded studies may hire a computer programmer to create personalized normative feedback and risk feedback graphics, or researchers may use existing online programs such as eCHECKUP TO GO/E-Chug (e.g., Palfai et al., 2014; Thompson et al., 2018) or gamified online personalized normative feedback (e.g., LaBrie et al., 2019). However, for research teams that do not have sufficient funding, these options may not be feasible. Thus, there is a crucial need for researchers, interventionists, and prevention programs to have access to an easily accessible and cost-effective template to create personalized normative feedback and risk feedback graphics using semi-automated methods.

With this goal in mind, the current article describes the process of developing and implementing a semi-automated and comprehensive personalized normative feedback and risk feedback syntax file for displaying personalized normative feedback and risk feedback. We outline a semi-automated approach we developed for presenting personalized normative feedback and risk feedback graphics in the context of a prevention program targeting alcohol use, risky sexual behavior, and sexual aggression in college-aged men. The approach uses IBM Statistical Package for Social Sciences (SPSS; IBM Corp., Armonk, NY) and Microsoft Excel/PowerPoint, which are available at low or no cost to universities, to create feedback graphics. In the remainder of this article, we describe our process of creating personalized normative feedback/risk feedback graphics, discuss lessons learned, and provide documentation for researchers and interventionists with limited resources seeking personalized normative feedback and risk feedback interventions that provide complex feedback.

Current project

The current project used personalized normative feedback and risk feedback as part of a prevention program designed to simultaneously reduce heavy drinking, risky sexual behavior, and sexual aggression. Participants first completed an online battery of questionnaires (to provide information for feedback) and then attended two laboratory sessions in which they received a personalized normative feedback/risk feedback intervention in addition to cognitive training or treatment as usual. The cognitive training modules provided participants with training regarding cues of sexual interest and/or disinterest and had participants practice identifying these cues in simulated experiences with potential partners (e.g., Treat et al., 2016, 2017).

Participants received personalized normative feedback on various attitudes, behaviors, and cognitive outcomes. However, the current tutorial focuses on the following: binge drinking frequency, risky sexual behavior, sexually aggressive behavior, and rape-supportive attitudes. Participants also received risk feedback on these variables (i.e., presenting item-level descriptive information in addition to the normative feedback) in addition to information about fraternity/sorority and intercollegiate athletics membership, impulsivity/sensation seeking, bar/party attendance, acceptability of heavy drinking, family history of alcohol use disorder, positive alcohol expectancies, weekly use of pornography, and misperceptions of sexual interest.

The current manual may be most helpful for researchers conducting in-person or mailed personalized normative feedback interventions that incorporate complex constructs that require data transformation (e.g., attitudes, expectancies) and incorporate other components outside of personalized feedback (e.g., cognitive training) that may necessitate other programs than a singular web platform.

Before reading the following sections, where we outline the steps to create personalized normative feedback and risk feedback graphics, we recommend opening both the SPSS syntax file and the Excel worksheet (see the supplemental material). (Supplemental material appears as an online-only addendum to this article on the journal's website.) Having both open while reading the manuscript will allow readers to navigate to specific lines of syntax in the SPSS syntax file and cells in the Excel worksheet referenced in the current tutorial.

Creating personalized normative feedback/risk feedback graphics

Step 1: Creating the IBM SPSS syntax file. For demonstrative purposes, we use fictitious data entered into a Qualtrics survey by the lead author to reflect a “high-risk” individual (i.e., participant ID #1). Although we used Qualtrics for survey data collection in our project, the process we outline is appropriate for any data collection platform (e.g., SurveyMonkey, REDCap) that provides exportable data files in common formats (e.g., SPSS or XLS). Once the data are collected, they can be saved as an SPSS file (.sav) or saved as an Excel file and imported into SPSS to support the presentation of feedback, regardless of the data collection source. We used IBM SPSS for data processing and transformation and then imported a data file that had only variables of interest (i.e., singular items, transformed sum/average scores) into Microsoft Excel and PowerPoint for graphics (see Table 1 for an overview of steps in the process of generating feedback graphics).

Table 1.

Overview of steps to create personalized feedback graphics

SPSS syntax steps	Specific instructions
Download data and import it to SPSS	Data collected should be saved as a .sav or .csv file and imported into IBM SPSS.
Clean and transform data	Data should be cleaned and transformed (e.g., aggregate scores) as necessary.
Label data to be used for personalized normative feedback	Give each individual data point (e.g., binge drinking frequency) or aggregate score (e.g., sexually coercive attitudes) a variable label.
Restructure data to be used for personalized risk feedback	One-by-one, transform each variable into cases, indicate which variables were endorsed by participants, create a random sample of items endorsed as yes/no or at a specific threshold, then convert the quantitative score into the text of that item.
Specify participant number	Specify for which participant you are seeking to create feedback graphics.
Keep only relevant variables and save into an Excel sheet	Delete nonrelevant variables from the syntax file and save the data for the participant into an Excel worksheet.
Excel worksheet steps
Create bar graphs for personalized normative feedback	Use the graph creator and select the participant's individual response for a construct (e.g., binge drinking frequency), their perception of peers, and their actual peer's behavior for each bar in the graph.
Create tables for personalized risk feedback	Use the table creator to select responses on items the participant endorsed, loop these items into a table, and give descriptions of the item and/or threshold of what is considered “high risk.”
PowerPoint presentation steps
Copy and paste special	Select the graphic from the Excel worksheet that you want to bring into PowerPoint, right click on the graphic and then click “copy.” Click on “Paste Special” in PowerPoint and select “Embed Using Source Formatting,” which brings graphics into the PowerPoint while embedding data to be changed each time the Excel sheet is updated.
Save individual PowerPoints for each person	Using unique file names for each participant will ensure that data from a different participant are not updated into an older participant's individualized PowerPoint.

Note: The current process used a combination of Qualtrics for data collection, IBM SPSS, Microsoft Excel, and Microsoft PowerPoint.

The first step in creating the syntax file was making sure that all data were cleaned and transformed as necessary (e.g., recoding variables to the scale of interest, reverse-scoring items, sum/average scoring of continuous constructs [e.g., risky sexual behavior, sexually coercive attitudes]; syntax lines 9–88). For items that did not necessitate sum scores (e.g., binge drinking frequency [single item]), the syntax file kept each person's individual score (e.g., self-reported binge drinking frequency) and each person's normative variable (e.g., self-reported perceived peers' binge drinking frequency). For items that necessitated data transformation (e.g., average sexually coercive attitudes out of 20 items), this syntax file also created sum/average scores for each person (e.g., average sexually coercive attitudes) and each person's normative variable (e.g., perceived peers' average sexually coercive attitudes) for personalized normative feedback. It is important to note that the current study did not have missing data on items that created higher-order sum/average scores. However, if other studies observe missing data, the current best practice is to create average/sum scores based on the proportion of items available (see Enders, 2023; Johnson, 2013).

Actual normative data came from past studies and was not a part of this data collection. After sum/average scores were created for continuous variables encompassing several items (e.g., sexually coercive attitudes), the syntax file individually reformatted item-level variables (e.g., specific sexually coercive attitudes) for risk feedback one by one. This process allowed individual items to be looped into the feedback and referenced if they were endorsed (i.e., yes/no) or endorsed at a specific threshold (relative to a normative sample).

For risk feedback, we use alcohol expectancies and rape-supportive attitudes as an example. For expectancies, our goal was to generate graphics that would display items endorsed by participants as “slightly agree” or “agree” to stimulate introspection about the items endorsed and how they may relate to their drinking and sexual behavior. Further, we sought to display a random selection of items that were endorsed as “slightly agree” or “agree” to (a) mitigate biases based on displaying certain expectancies and (b) include expectancy items that may appear at the end of the scale but may be just as important as items appearing at the beginning, rather than simply displaying the first several items on the scale.

To achieve this, we generated the following: (a) a specific number of items at a specific threshold for each construct (e.g., four positive expectancies that were rated at 4+/5 on a scale of 1–5), (b) a random sample of items that met the threshold (e.g., if a participant had six expectancies endorsed at 4+, we wanted a random sample of three of them), and (c) the name of each item to replace the score (e.g., we wanted a score of 4/5 on Expectancy Item #4 to be converted to a text response for that item [e.g., “I will feel sociable”]). To achieve this, each variable had to be reformatted into a case.

VARSTOCASES/MAKE CEOASOC FROM CEOA_1 CEOA_3 CEOA_5 CEOA_14 CEOA_24 CEOA_31 CEOA_34 CEOA_38/INDEX=SocExp(CEOASOC)/KEEP=ALL/NULL=KEEP. alter type SocExp(a200). RECODE SocExp (‘CEOA_1’=‘I would be outgoing’) (‘CEOA_3’=‘I would be humorous’) (‘CEOA_5’=‘It ‘+ ‘would be easier to express my feelings’) (‘CEOA_14’=‘I would be friendly’) (‘CEOA_24’=‘I ‘+ ‘would feel energetic’) (‘CEOA_31’=‘t would be easier to talk to people’) (‘CEOA_34’=‘I would ‘+ ‘be talkative’) (‘CEOA_38’=‘I would act sociable’). EXECUTE.

[Syntax Lines 267-277/309-322]

We then created syntax to (a) select all items within a construct that met the threshold, (b) randomly select a set number of these items for presentation, and (c) recode numbers/items into words. To ensure that the ordering of items stayed exactly the same for every participant, we had to ensure that “expectancy 1”, “expectancy 2”, “expectancy 3,” and “expectancy 4” were in the same cells for each participant in the Excel spreadsheet. However, if one participant had seven expectancies above the threshold and another participant only had one, the person with only one expectancy would have three cells missing in the Excel spreadsheet since expectancy 2, expectancy 3, and expectancy 4 would be empty because of non-endorsement. Thus, we created syntax that (a) created a sum variable of the number of expectancies endorsed above threshold (or any construct of interest; see syntax lines 281–294 [.sps], 326–339 [.pdf]), (b) ordered the expectancies so that endorsed items would be at the top of the file (see syntax line 298 [.sps], 343 [.pdf]), and finally (c) deleted cases below the number of items selected (i.e., 4 expectancies; see syntax lines 302–327 [.sps]. 348–363 [.pdf]). Thus, if a person only had one expectancy above the threshold, the SPSS file saved an additional three expectancies, but the sum score indicated that only the first (sum score = 1) expectancy item was actually endorsed and, therefore, presented in the feedback.

USE ALL. COMPUTE filter_$=(CEOASoc >=3). VARIABLE LABELS filter_$ ‘SocAtt =1 (FILTER)’. VALUE LABELS filter_$ 0 ‘Not Selected’ 1 ‘Selected’. FORMATS filter_$ (f1.0).

FILTER BY filter_$. EXECUTE. AGGREGATE/OUTFILE=* MODE=ADDVARIABLES/BREAK=ID/CEOASoc_sum=SUM(filter_$).

[Syntax Lines 281-294/326-339]

SORT CASES filter_$ (d).

[Syntax Line 298/343]

USE ALL. do if $casenum=1. compute #s_$_1=1. compute #s_$_2=mean.(CEOASoc_sum).

end if. do if #s_$_2 > 0. compute filter_$=uniform(1)* #s_$_2 < #s_$_1. compute #s_$_1=#s_$_1 - filter_$. compute #s_$_2=#s_$_2 - 1. else. compute filter_$=0. end if.

VARIABLE LABELS filter_$ ‘1 from the first XXX cases (SAMPLE)’. FORMATS filter_$ (f1.0). FILTER BY filter_$. EXECUTE. SORT CASES filter_$ (d). FILTER OFF. USE 1 thru 1/permanent.EXECUTE.

[Syntax Lines 302-327/348-363]

The same process was used for rape-supportive attitudes, where we sought to provide graphics with feedback on attitudes that participants endorsed any agreement with. Thus, the threshold was set to 2, being anything more than “not at all agree.” Two endorsed rape-supportive attitudes were then displayed, using the same procedures as for expectancies. Once this process was completed for all variables of interest, we dropped nonrelevant variables in SPSS, recoded missing variables to zero, and saved the file as an Excel document (.XLXV; see SPSS syntax lines 505–514 [.sps], 553–565 [.pdf]).

SAVE TRANSLATE OUTFILE=‘/Users/jtwaddell/Dropbox (ASU)/preventiongrantID12.xlsx’/TYPE=XLS/VERSION=12/MAP/FIELDNAMES VALUE=NAMES/KEEP = ID Greek Sports Party Bar Porn Misperception FamilyHistory BingeFreq BingeFreqNorm Accept_Freq Accept_Quant UPPS_Mean UPPS_Sens SES_Total SES_Norms_Total IRMA_Mean IRMA_Norms_Mean RiskySex_Mean RiskySex_Norms_Mean RapeSupp_sum RiskySex_sum RSUPP.1 RSUPP.2 RSUPP.3 RSex.1 RSex.2 SocExp CEOASoc_sum SexExp CEOASex_sum LCExp CEOALC_sum TRExp CEOATR_sum/CELLS=VALUES.

[Syntax Lines 505-514/553-565]

Once this was completed, lab personnel could input the participant ID number into the syntax to delete data from other participants. Then, the lab personnel could highlight the full syntax file, hit run, and generate and save an Excel file with data on all variables of interest.

Step 2: Creating the Microsoft Excel feedback document. The current study used Excel (looped into PowerPoint) to present personalized normative feedback/risk feedback. We first saved data from SPSS into an automatically savable Excel file so that it could be directly imported into the final Excel file where equations and graphics used are housed. For demonstrative purposes, we took data from the fictitious participant and used the saved Excel file. Within this file, we created bar graphs for personalized normative feedback and tables for risk feedback. For personalized normative feedback, we used the graph creator and “select data” to create the graphics. For example, to create the binge drinking graphic in Figure 1, we selected the data in Column I (row 2 = participant response, row 4 = normative mean stable across participants) and Column J (row 2). For risk feedback, we wanted tables that indicated (a) whether an individual was at high/low risk, (b) the individual's exact level of risk, and (c) specific items endorsed in the relevant domain. To do this, we used Excel equations within each table to display whether a person was at high/low risk based on how their values on certain constructs compare to normative values (e.g., impulsive traits). All equations are shown in the supplemental file “ExcelPNFRF.”

Figure 1.

Personalized normative feedback graphics

Although the current study had a variety of feedback graphics, the “overall risk feedback” table (Figure 2; Excel Columns AA–-AC5–15) was presented as text. To create this table, we first typed each feedback target (e.g., “Impulsivity”) into the first column of the table. Next to each column, an equation was used to indicate if the individual endorsed the item above the threshold (e.g., Participant mean impulsivity = 2.5, Threshold of impulsivity = 2.08). If the value exceeded the threshold, the text displayed “High,” and if the value did not exceed the threshold, the text displayed “Low.” In addition to displaying this text, the next line was programmed such that if the text read as “High,” Excel displayed “Above average,” and if the text read as “Low,” Excel displayed “Below average.” This approach was used for Overall, Alcohol, and Sexual Aggression Risk Feedback (Figures 2, 3, and 4; Excel Columns AA–AC5–28). It is important to note that all fonts/positioning/text sizing, etc., were editable via this process to cater to each research team's needs, which may be another feature that is not available via Qualtrics.

Figure 2.

Global risk feedback

Figure 3.

Alcohol-specific risk feedback

Figure 4.

Sexual aggression–specific feedback

In addition to providing information about risk status, we wanted to display the exact wording of individual items that participants endorsed. To do this, we created tables displaying items in each domain up to a specified randomly selected number of items. Tables for alcohol expectancies (Figure 5) and rape-supportive attitudes (Figure 6) are described here. For the expectancies table, we first created a title and header (similar to before) indicating that the items were positive expectancies that the participant endorsed. After the header was created, each subsequent line had an equation that specified randomly selected positive expectancies endorsed (or erroneously included if none were endorsed; see Excel Columns AE13–16). In addition, we had to specify the number of total expectancies endorsed above the threshold (described above; see Excel columns AC2, AE2, AG2, and AI2) so that expectancies would be displayed up to the number endorsed. Thus, using the aggregate score (of how many positive expectancies were endorsed from each subscale of expectancies [sociability, sexuality, liquid courage, tension reduction]), each line of syntax specified the inclusion of the positive expectancy if the expectancy sum was greater than or equal to 1 for that line. Alternatively stated, the equation in the first line of the table (AE13) says, “If the participant reported more than one sociability expectancy above the threshold, display the first item; if not, display a blank space”; lines 2 (sexuality expectancies; AE14), 3 (liquid courage expectancies; AE15), and 4 (tension reduction expectancies; AE16) of the table were constructed the same way.

Figure 5.

Risk feedback on alcohol expectancies

Figure 6.

Rape-supportive attitudes risk feedback

Similar to expectancies, risk feedback for rape-supportive attitudes that were endorsed at a threshold of 2 or above (i.e., 1 = not at all agree) was also displayed. For the rapesupportive attitudes table, we first created a title and header (similar to before) indicating that the items were “sexually coercive attitudes” that the participant endorsed. After the header was created, each subsequent line had an equation that specified randomly selected rape-supportive attitudes endorsed (or erroneously included if none were endorsed; see Excel Column V2). In addition, we had to specify the number of total attitudes endorsed above the threshold (described above) so that rape-supportive attitudes would be displayed up to the number endorsed. Thus, using the aggregate score (of how many rape-supportive attitudes were endorsed), each line of syntax specified the inclusion of the rape-supportive attitude if the attitude sum was greater than or equal to 1 for that line (see syntax lines 144–159). Alternatively stated, the line 1 equation says, “If the participant reported more than one rape-supportive attitude, display the first item; if not, display a blank space”; Lines 2 and 3 were constructed identically.

Step 3: Creating the PowerPoint file. The final step was to import the graphics from Excel into PowerPoint. We found that this must be done on a Windows computer, as it is not possible to embed Excel files into PowerPoint and have the graphics automatically change as the values in Excel change using a Macintosh computer. We copied each graphic in Excel (right click on graphic and then click “copy”), clicked on “paste special” in PowerPoint, and then selected “Embed using source formatting” to bring each graphic into the PowerPoint, while embedding data. Once this was done, the PowerPoint file was saved and used for all participants; all edits to create person-specific feedback happened in the Excel file, which automatically updates the PowerPoint figures for viewing (see Figure 7 for an example PowerPoint slide).

Figure 7.

Example PowerPoint slide

As new cases were added to the Qualtrics data file that is extracted for normative feedback, the SPSS syntax file was used for each person's data. For example, if a second participant was added to our fictitious data set and the updated SPSS data file was saved in Excel form, the second row of the Excel file was copied and then directly pasted into the working Excel file that contained the tables and graphics. When the new participant's data were pasted into the working Excel file, the equations reformulated graphics for the new participant, which were already embedded into the PowerPoint file. Note that each person's Excel file and PowerPoint file must be saved as a unique file with the participant ID (e.g., ExcelFile_P1, PP_P1).

Lessons learned

The approach outlined above is relatively time intensive because the syntax file for the larger grant was much longer than the example file included here. Although the investment on the front end is considerable, this approach allows individuals with relatively little training and resources to create personalized normative feedback and risk feedback for each new participant, reducing staff burden during the trial.

In addition, it is important to ensure that feedback will be appropriate for all participants in the trial. Thus, after the steps outlined in this article have been completed, we strongly recommend generating fictitious data for participants with a variety of risk profiles to ensure that the feedback displays are appropriate for individuals with a broad range of attitudes and behaviors.

Finally, Excel will automatically set values for the x-axis when creating graphics, but these values may not be theoretically and/or visually appropriate. Thus, these values can be modified by the researcher. It is important to ensure that all axes are displaying the data appropriately before finalizing the Excel document that loops into the PowerPoint document. Also, it is important that each file be given a unique name when saving each person's individual Excel and PowerPoint files. This reduces the likelihood that the final versions of the Excel and PowerPoint get written over and that each participant is shown their feedback, not someone else's.

Adaptability to other projects

We believe the current process for creating personalized normative feedback and risk feedback graphics could translate to nearly any construct. Since we present (and describe the process for) single-item feedback, sum/average-score feedback, and text-based feedback, the current process should be adaptable to most other studies. When adapting the current process, it may be particularly important to check and ensure that each section of the SPSS syntax is doing what is intended, as one missed step could affect the whole syntax file. It is also particularly important to ensure that the equations used in the Excel worksheet translate to other data types (e.g., data from platforms other than Qualtrics, nominal/ordinal data that are not continuous). It is likely that the equations can be modified to suit data that are substantially different than those used in the current study. For instance, instead of providing risk feedback about an individual's frequency of bar and party attendance, a researcher could program Excel equations to display protective strategies if one does attend bars and parties frequently. In summary, the process outlined here should be applicable to any type of feedback, but researchers should meticulously check and extend the current process when adapting it to their own studies.

Conclusions and future directions

Personalized normative feedback interventions and provision of risk feedback as part of brief interventions are commonly used to target health risk behaviors and have been shown to be effective at reducing some forms of risky behavior, particularly among college students and emerging adults. Implementing personalized feedback interventions can be challenging, particularly with respect to creating feedback graphs and displaying the graphs in a way that is accessible and visually appealing. In fact, the complexity of designing, implementing, and displaying personalized feedback could be a barrier to large-scale dissemination and implementation of personalized feedback if it is not already pre-packaged in a web module (e.g., eCHECKUP TO GO) or if substantial grant funding isn't available to hire programmers. The current study provides scripts and syntax to help researchers, interventionists, and prevention program developers automate several steps with their own data. The authors hope that the current tutorial will aid other researchers in conducting this work without human error or substantial funding to support their work and will aid other researchers in thinking through the steps/methods of personalized feedback and best practices for generating personalized graphics.