Estimating Utilities with Latent Class

Top  Previous  Next

Note: Latent Class is often used to discover market segments (for use as banner points) from CBC data.  But, Latent Class is not used nearly as often as HB among our customers for producing the final utilities that are used when developing market simulators.

Lighthouse Studio includes a built-in Latent Class estimation routine.  You may also decide to analyze data separately using our standalone Latent Class software.

When you run Latent Class (Analysis | Analysis Manager...  then by selecting Latent Class as the Analysis Type then by clicking Run), the results are displayed in the report window and utilities are saved into a subfolder within your project directory.  You can weight the data, select subsets of respondents or tasks to process.


Latent Class is a utility estimation method for use with CBC (Choice-Based Conjoint) or MaxDiff data.  It assigns respondents into segments having similar preferences based on their choices in CBC questionnaires.  It uses latent class analysis for this purpose, which simultaneously estimates part worth utilities for each segment and the probability that each respondent belongs to each segment.  Latent Class is an integrated analytical component within Lighthouse Studio (and we also provide a standalone Latent Class software system).  Results from latent class runs plus segment membership as filters/banner points can be taken forward into market simulations.


Latent class has a role analogous to that of CBC's logit program, but rather than finding average part worth utilities for all respondents together, it detects subgroups with differing preferences and estimates part worths for each segment.  The subgroups have the characteristic that the respondents within each group are relatively similar but the preferences are quite different from group to group. You may specify how many groups are to be considered, such as the range of 2 through 6.


The latent class estimation process works like this:


1. Initially, select random estimates of each group's utility values.
2.Use each group's estimated utilities to fit each respondent's data, and estimate the relative probability of each respondent belonging to each group.
3. Using those probabilities as weights, re-estimate the logit weights for each group.  Accumulate the log-likelihood over all groups.
4.Continue repeating steps 2 and 3 until the log-likelihood fails to improve by more than some small amount (the convergence limit).  Each iteration consists of a repetition of steps 2 and 3.


Latent class reports the part worth utilities for each subgroup or "segment."   Latent class analysis does not assume that each respondent is "in" one group or another.  Rather, each respondent is considered to have some non-zero probability of belonging to each group.  If the solution fits the data very well, then those probabilities approach zero or one.


Use of latent class analysis as a segmentation method has been examined in many articles in the marketing literature, and it has been found to be effective for that purpose.  Thus, it holds promise of solving the problems occurring with aggregate conjoint analysis:


There is usually sparse information at the individual level for defining segments.

If there truly are segments with different preferences, an aggregate analysis may give incorrect answers.


Latent class became popular in about the mid-1990s as a tool for analyzing CBC data sets.  The model typically provided more insight about the structure of respondent preferences than aggregate logit, and the resulting market simulations were usually more accurate than similarly defined aggregate models.  The latent class approach was effective in reducing the negative effects of the IIA assumption in logit analysis.


For more information on our latent class analysis approach, please see the full standalone Latent Class manual.

Page link: