Transportation Security Index

The TSI was designed to identify individuals experiencing transportation insecurity. It is only able to do so when (1) all the questions in the index are administered, (2) each of the questions are individually scored, and (3) the scores of each of the individual questions are added together, thus producing the respondent’s individual sum score (see “How do I score the TSI?” for instructions on scoring).

To better understand how the TSI works as an index, let’s use an analogy: diagnosing the flu. If you go to the doctor they usually ask you what symptoms you are experiencing: Do you have a fever? Do you have a runny nose? Are you coughing? Do you have the chills? Do your muscles ache? On their own, none of these symptoms help the doctor diagnose if you have the flu versus something else, like the common cold. Considered together, however, the doctor is better able to make this assessment. 

The TSI works similarly. Only when all the questions in the index are analyzed together, can the index be used to “diagnose” if someone is experiencing transportation insecurity. 

Each version of the TSI performs similarly well in terms of their ability to identify respondents experiencing transportation insecurity and the prevalence estimates they generate. The biggest differences between the versions is how many categories of transportation insecurity they can identify and their time to completion.

For those who simply want to distinguish between people who are transportation secure vs. those experiencing insecurity, the TSI-3 offers the ability to do so in an efficient, cost effective way.

There are times, however, where being able to distinguish between different categories of insecurity are useful and worth the tradeoffs of using a version of the TSI that takes longer to complete. For example, when used to evaluate an intervention, it can be useful to know if the intervention is moving people from high insecurity to low insecurity. Even if the overall percentage of people are still insecure after intervention implementation, alleviating the severity of insecurity might be deemed a success and have other positive impacts, something difficult to see with just two categories. Similarly, in research, users may want to know how people with different levels of insecurity differ from one another or may want to better understand what is driving these different levels of insecurity. Such information can help people identify the different kinds of solutions that will alleviate these different kinds of transportation insecurity.

Notably, all versions can be used as a continuous variable.

To score the TSI, look at the responses to each of the individual questions in the TSI. Assign each “never” response a score of 0, each “sometimes” response a score of 1, and each “often” response a score of 2.

To identify a respondent’s transportation insecurity score, sum the scores of each of the individual question responses. To see how to use this score to assign a respondent to a transportation insecurity category, please download the instructions for the version of the TSI you are interested in using.

Questions in the TSI reflect the material and relational symptoms of transportation insecurity as observed in qualitative research. The material symptoms reflect the difficulties people have getting from place to place in a safe or timely manner (e.g., arriving places late, skipping trips, rescheduling appointments). The relational symptoms reflect the emotional toll and social strain of experiencing transportation insecurity (e.g., being embarrassed, feeling left out, worrying about inconveniencing ride givers). If your goal is to measure transportation insecurity, all questions in the index must be used (see “How do I measure transportation insecurity using the TSI?” for more information). 

While many people are familiar with the material symptoms of transportation insecurity, our team wishes to emphasize the importance of the relational questions in the index. Our qualitative research uncovered that the relational symptoms of transportation insecurity were a key way transportation insecurity manifested in the everyday lives of people and could be just as troubling for people as the material symptoms. Our quantitative analysis has confirmed this. Transportation insecurity is a unidimensional condition experienced both materially and relationally. To measure transportation insecurity thus requires measuring both types of symptoms.

Unfortunately, no. The Transportation Security Index was developed using complex, rigorous statistical methods to determine how questions are being endorsed and how they relate to one another in ways that constitute an index that measures the latent construct of interest (i.e. transportation insecurity). Accordingly, while you may use a subset of questions that comprise the index in your own work, it is important to recognize that no subset of questions that are different from those validated in the three versions of the TSI will identify transportation insecurity itself (for more, see “How do I measure transportation insecurity using the TSI?”). Nor will a subset of questions that are different from those validated in the three versions of the TSI identify transportation insecurity itself. 

We strongly advise against this. In terms of changing item wording, the construction of each item was informed by extensive qualitative research and refined using cognitive interviews. Through such interviews, we know that respondents understand the questions as they are phrased and that their responses are capturing the kind of data that is intended. Altering question wording could compromise respondent burden, comprehension, and judgement. Furthermore, the questions as they exist are those that have been validated; altering their wording may change the results in ways that impact the TSI’s ability to measure transportation insecurity. The same is true of changing the ordering of questions. The ordering as they are presented is the ordering that has been validated. Changing the ordering (or adding additional questions in between those items that comprise the index) may change the results in ways that compromise the quality of the data and the ability to make comparisons across findings.

Certainly! While measuring transportation insecurity requires users to analyze all of the questions in the index together (see “How do I measure transportation insecurity using the TSI?”), users can also analyze responses to each of the individual questions in the TSI to better understand how people are experiencing each of the symptoms of insecurity. Doing so will not tell you if the respondent is transportation insecure but it will provide you with a descriptive portrait of how your respondents are experiencing each of the symptoms of transportation insecurity. For an example of how the TSI has been used as an index to generate information about transportation insecurity and how each of the questions have been analyzed to create the descriptive portrait, see “Transportation Insecurity in the Motor City.”

At the current moment, the TSI has only been validated in English. However, there are researchers who have done their own translations of the TSI in different languages. Some of these researchers are working to validate these translated versions of the TSI. If you would like to know more or would like to work with our team to translate the TSI, please let us know. 

The TSI was developed and validated in the United States. Because it measures transportation insecurity through questions that tap into “symptoms” of this condition, it is possible that the symptoms observed in the U.S. may not be found elsewhere. For instance, in certain cultures and contexts people may not worry about inconveniencing ride givers because car pooling or ride giving may be more of a cultural norm than in the U.S. We can not know how the TSI will translate to non-U.S. contexts without testing and validation in these contexts. That being said, it is being deployed outside of the United States: in Australia. If you would like to work with our team to do testing and validation outside of the U.S., please let us know.

Contact us! We are here to help.

Gould-Werth, Alix, Jamie Griffin, and Alexandra K. Murphy. 2018. “Defining a New Measure of Transportation Insecurity: An Exploratory Factor Analysis.” Survey Practice, 11:2, pp. 1-36.

Murphy, Alexandra K, Alix Gould-Werth, and Jamie Griffin. 2021. “Validating the Sixteen Item Transportation Security Index in a Nationally Representative Sample: A Confirmatory Factor Analysis” Survey Practice. 

MacDonald-Lopez, Karina, Alexandra K. Murphy, Jamie Griffin, Michael Bader, Alix Gould-Werth, and Nicole Kovski. 2023. “A Driver in Health Outcomes: Developing Discrete Categories of Transportation Insecurity.” American Journal of Epidemiology.

Murphy, Alexandra K., Alix Gould-Werth, & Jamie Griffin. 2024. “Using a Split-Ballot Design to Validate an Abbreviated Categorical Measurement Scale: An Illustration Using the Transportation Security Index.” Survey Practice, 1-17.