Yes, it does! Not only do temperature spikes (bad data) affect averages, when the number of stations recording data is small, the effect is amplified to give higher readings on average all around.
If stations are closed, IIRC, the practice was to average data from adjacent stations to fill in the dataset. With averages similarly corrupted by questionable data, the overall picture would reflect a trend which did not, in fact, exist.
Thanks for posting this!
Russia's top weatherman's .....says winter in Siberia may be COLDEST on record