### Q 1: When is it appropriate to use PCI_{2}?

A: PCI_{2} can be used with scale widths of 2, 3, 4, 5, 6, 7, 8, and 9, and can be applied to bipolar scales (with or without a neutral value) and unipolar scales (e.g., not at all important to extremely important). It is not appropriate for truly continuous variables such as number of days of participation in an activity (coded as 0, 1, 2, 3, … 365 days).

The statistic is called the Potential for Conflict Index but is not limited to conflict evaluations. PCI_{2} can be used in any situation where the concern is estimating the amount of consensus that exists regarding variables of interest; either for the population or sample as a whole or for specific segments (e.g., males vs. females).

### Q 2: Which version of PCI_{2} should I use – SPSS, SAS, Microsoft Excel,

or the standalone version (PCI_{2sa})?

A: All versions of PCI_{2} calculate the statistic in the same way and current testing indicates that all versions of PCI_{2} yield the same results. SPSS and SAS, however, are relatively expensive statistical software applications that not all users have access to. Even among those who might have SPSS, novice users may not feel comfortable using the macro. Similarly, not all users have Microsoft Office and its suite of applications (e.g., Excel). The standalone version (PCI_{2sa}) was developed to accommodate users who may not have access to the commercially available software. Individuals should select whatever version they feel most comfortable with.

### Q 3: Are there any sample size restrictions when using the Excel version?

A: Versions of Excel before 2007 are limited to a sample size of approximately 65,000. Use PCI_{2} with Excel 2007 or greater if you have more than 65K observations.

### Q 4: Can I use weighted data with PCI_{2}?

A: We recommend using unweighted data with PCI_{2}. Future releases of PCI may accomodate the use of weighted data.

### Q 5: I typically code my bipolar scales as 1 (Strongly disagree) to 7 (Strongly agree). Do I need to recode these variables to -3 to +3?

A: Yes, we recommend recoding the variables to -3 to +3. You need this coding scheme to calculate the correct scale mean when graphing the variables.

### Q 6: What distance function should I use?

A: The answer partially depends on the type of scale being analyzed; for example:

For bipolar scales with a neutral value you can use either

D_{1} (excludes neutral value from distance calculations) or

D_{2} (includes neutral value in distance calculations)

For bipolar scales without a neutral value use D_{1}

For unipolar scales use D_{3}

A: For bipolar scales with a neutral value the choice of D_{1} or D_{2} depends on whether there is reason to believe that the neutral value should be included. We currently recommend using D_{1}, but encourage researchers to experiment with both alternatives.

### Q 7: What does the power function do?

A: Distance functions need not be linear functions of responses. Powers of differences reflect non-linear perceptions of distance. For example, someone with +1 may not see someone responding with -1 as being much in conflict. Someone responding -3, however, may be seen as threatening to what the +1 person wants because the -3 person may push strongly for change. To reflect these non-linear perceptions, the difference scores can be raised to some power. If the initial difference scores were 1, 2, and 3 (i.e., power = 1), squaring the differences (i.e., power = 2) results in distances of 1, 4 and 9. A power of 2 gives more weight to larger differences between individuals. The greatest difference occurs between individuals who express the most extreme values on a scale (e.g., for a 7-point scale for D_{1} -3 and +3 differ by 36). The PCI_{2}estimation allows for alternative powers (e.g., 1, 1.5, 2) greater than 0 and less than 5.

We currently recommend using power = 1, but encourage researchers to explore with other power functions.

### Q 8: Why do I need to run the simulation?

A: Simulation is used to measure the observed PCI’s variability in an inferential setting. The simulation amounts to creating new survey data of the same sample size as the original data by repeatedly sampling with replacement from the observed data; 400 samples of the observed sample size are generated. The default of 400 can be changed to any sample size. The analysis produces a simulated PCI_{2} mean and standard deviation. The standard deviation of the simulated PCI_{2} is an approximate standard error and is critical in testing for a significant difference from a value (e.g., from 0 or 1) and for testing for significant differences among PCI_{2} values.