--multicategory_normalization

Switch

--multicategory_normalization

Description

Like --lexicon_normalization, but normalizes each group's category values by the total lexicon-term count across all categories combined, rather than per category. Aliased as --liwc_normalization, --liwc_norm.

Argument and Default Value

No argument. Boolean switch, default: False. Verified: implicitly turns on --lexicon_normalization too (if args.multicategory_normalization and not args.lexicon_normalization: args.lexicon_normalization = True), so it always implies that switch.

Details

Verified: with a real weighted lexicon, every group's group_norm summed to almost exactly 2.0 across all categories combined (vs. ~26–33 for --lexicon_normalization alone, and ~1.2–1.3 with neither switch) — a much more uniform scale, useful when a message can belong to more than one category (e.g. LIWC) and you want category values to be comparable across groups regardless of how lexicon-heavy their language was overall.

Other Switches

Required Switches:

• --add_lex_table

• -l

Example Commands

Using the packaged dla_tutorial dataset (see Differential Language Analysis (DLA) Tutorial):

dlatkInterface.py -d dla_tutorial -t msgs -c user_id --add_lex_table -l met_a30_2000_cp \
--weighted_lexicon --multicategory_normalization