Quantitative Research

Quantitative Research

In looking for a healthcare provider, customers are searching for not only affordable and great care, but also a memorable experience. A mundane logo can take away from such an experience. In general, healthcare logos are blue or green in color to symbolize calmness along with wellness. Intermountain Healthcare’s logo fits this mold which can make it difficult to stand out in such a competitive environment. Although the current logo is steps ahead of past logos, it remains almost entirely meaningless. It has very little to do with health care at all and makes it difficult for customers to associate it with the product, which is very problematic. Intermountain Healthcare may work toward developing more meaningful and memorable for customers by utilizing a 2x2 research method. In this research we studied the effect of 2 variables, new logo and new color scheme, have on customers’ thoughts of trust, comfort, and compatibility with the company. we launched a survey across the U.S. to gain insights on what people outside of the Intermountain area may think about our logos and then processed the data to evaluate our hypothesis.

Team: 4 people
Time Span: 1 month

My Role

Questionnaire Designer & Distributor
Data Analyst
Supplementary Factor Researcher & Interpreter

What Did We Try To Learn?

Will a change in shape or color influence the reaction that consumers have with a healthcare system?

In order to understand this problem, we designed a 2×2 study along with a survey, in Qualtrics, that randomly shows one logo in each independent variable respectively, so the participants will see two of the four logos. Three new logos were designed and the current logo was used as the control. We asked participants to score the logos using a 7-point scale based on trust, comfort, and compatibility. Covariate questions were also asked. The survey was launched on Amazon Mechanical Turk and collected at least 200 responses only in the United States.

Tools: Qualtrics, Amazon Mechanical Turk, Excel, JMP

Methods: questionnaire/survey, logo design, data cleaning, hypothesis tests (regression test, t-test and Anova test), data interpretation

Methods

(1)  Experimental design.

We designed a 2×2 study along with a survey, in Qualtrics, which randomly shows one logo in each independent variable respectively, so the participants will see one of the four logos. To follow the funnel analogy and avoid leading sequences, we order our questions from general to specific. First, we asked them to score the logos in a 7-point scale based on trustworthiness, quality, and reliability. Then we asked them covariate questions regarding times of visiting hospitals, reason rankings of choosing hospital and ratings of current hospital in different perspectives. We planned to implement the survey on Amazon Mechanical Turk and collect at least 200 responses only in United States. After we collect our data, we will go through data cleaning, exploration, analysis and interpretation with Excel and JMP. We will run hypothesis tests including regression, t-test and Anova to determine the significance of main effects, interaction effects and covariate effects. After analyzing the data, we will also be able to understand how the themes and colors of logos influence customers’ thoughts toward health care service.

(2)  Independent variables

For our tests, we designed three new logos: one is the original logo with new color, one is a new logo with the original color, and the last one is a new logo with a new color. We then used the current logo as a control.

Our first independent variable is the controlled logo with the different colors and our second independent variable is the controlled color with the different logos.

(3)  Dependent variable

Our dependent variables are the participants’ thoughts of intermountain healthcare regarding trust, quality, and reliability, which will be indicated with scores on a 7-point scale. We also conducted a multi-item scale to determine its reliability. Since the Cronbach’s alpha is 0.94 which is over 0.7, our multi-item scale is reliable.

(4)  Supplementary measures and covariates

We included times of visiting hospitals, reason rankings of choosing hospital and ratings of current hospital in different perspectives in covariates in order to understand what factors are important for customers in choosing hospitals. We also included gender, age range and education level so that we may know how participants’ thoughts vary based on these demographic factors.

(5)  Ingoing hypotheses.

We assume that the mean of the new logo will be significantly different from the original one; the mean of the new color will also be significantly different from the original color.

How Did We Prepare Our Data?

data cleaning

We collected our sample from participants in the United States through responses to a survey run on Qualtrics. The participants were exposed to the survey through Amazon Mechanical Turk. The participants of the survey received compensation of 10 cents for completing the survey which took about 3-5 minutes to complete. The survey was administered in 4 different versions at a randomized interval giving each independent variable to different people.

There were a total of 228 respondents that participated in the survey. Most of the respondents that were removed were due to incomplete surveys. There were also a few individuals that were removed because the image that we were testing did not show up on their screen. After removing the 35 participants that had given incorrect or irrelevant data we ended up with 193 participants. The 193 participants was the selection that we used to run the subsequent tests of significance.  We had a wide variety of demographics that were sampled. The majority of the individuals were from the 25-34 range with 92 total participants. We had an overall age range from 18-74. We had 82 male respondents and 118 Female respondents. Most of our respondents were well educated. 139 of our respondents had completed a 2 year degree or more. 29 had a professional degree and there were 3 doctorate recipients in the pool. In our sample 39% of people had completed a 4 year degree, which was the largest group in the education section.

What Have We Discovered?

data

With the results that we gathered, we found that there was significance when running a full factorial regression analysis. The participants who were shown the Intermountain health logo in the Control Shape responded more positively to the logo than the group that was shown the test shape. This response was significant (p= .024) among the scaled positive responses that we received from our dependent variables on perceived quality, trustworthiness, and reliability. The interaction effect of the Control Shape and Control Color were also significant (p=.0075) in the linear regression test that was run. While the main effect of color was not significant (p=.215) the interaction of keeping the control color had a larger effect on the control shape than the new shape that we were testing.

The main effect of the color was not significant in the test that we ran. There were two simple effects that were statistically significant. The most significant (p=.0002) was the difference between the control shape and new shape in terms of color. The mean of the control (mean= 5.413) was greater than the variable (mean= 4.673.) The group size tested in this situation was 100. The other simple effect that was significant (p= .003) was the control color and new color in terms of the control shape. The mean of the control color (mean= 5.414) was significantly higher than the variable mean (mean=4.826.) The group size in this test was 98.

What Else Have I Found?

Were There Limitations & Weaknesses In Our Research?

Internal validity ensures that there is a direct and understandable link between the cause and effect in a study. A common internal validity issue with these types of studies is that of improper sampling. With a local company, it is important that we refrain from simply “convenience sampling” or choosing respondents within our own geographical area because respondents’ all around Utah are likely already familiar with Intermountain Healthcare and their responses would likely skew results. It was for this reason that we decided to release our survey nationally to receive the best responses. Another common cause of internal validity issues is “confounding”. This means that outcomes may have been caused by something other than our main independent variable that we are specifically manipulating. An example of this is results skewed by feelings for hospitals in general, age, gender, and other demographic covariates. Specifically, we asked as a covariate how often our participants visited the hospital in an average year. We tested and found out that the number of visits per year does not significantly affect our dependent variable results. We also tested and discovered that although we had marginally more female respondents than male, gender did not significantly affect results. That is, male and female respondents have close to an equal amount of positive responses; females had an average response of 5 while males had an average response of 4.94.  We also tested to discover that respondents within the 55-64 year age range had the most positive responses (5.44) while respondents aged 45-54 had the lowest mean ratings (4.57). However, these results did not significantly affect the dependent variable results. It was interesting to discover, however, that education level significantly affected our dependent variable results We had 3 respondents with doctorate degrees and 1 that had less than a high school education. These turned out to be outliers and once we filtered out these results, our outcomes were more accurate.

External validity issues are defined as variables that cause the results to not be generalizable. For example, our study found that the blue and green logo color scheme generated significantly more positive results than did the orange and blue in our study. This does not necessarily mean that in any given setting, blue and green logos will always do better than orange ones. The way we set up our study did not lend the results to being overly generalizable in many settings other than our own.

What Did We Learn & Recommend?

For our tests, we designed three new logos: one is the original logo with new color, one is a new logo with the original color, and the last one is a new logo with a new color. We then used the current logo as a control. It is evident that after our thorough research process, it is suggested that Intermountain Health Care keep its current logo and color scheme. This combination significantly showed more positive reactions from respondents from across the nation in terms of satisfaction, quality, and care perception. There was a highly significant correlation between the logo shape and customer satisfaction. This means that there was a significant change in customer perceptions when they were switched between the control and the new logo shape. Color change also provided a significant change in customer satisfaction. This means that the main effect was significant while interaction effects were as well. However, it was not found that any covariates were significant, which means that our group is sure that our independent variables are the cause of the dependent variables of satisfaction, quality, and care. These results are actionable in the fact that Intermountain needs take no action and should keep their logo the same. No future research is necessarily needed at this time as the results were fairly conclusive.